On-Premise

When data must not leave your premises: software and AI on your own infrastructure.

dectria delivers software and AI applications for operation in your own infrastructure. Applications come as Docker containers that run on your own servers, in a private data center or in a hybrid environment.

For AI without the cloud, we use Ollama: open language models such as Llama, Mistral or Gemma run directly on your hardware, and requests never leave your network.

We support the whole path – from architecture and delivery to updates, backups and monitoring in operation.

Your contact Michael Jauk Contact for cloud & infrastructure

What is on-premises – and when is it worth it?

On-premises means that software runs on servers you operate or control yourself – in your own data center or at a hosting provider of your choice. If applications should not run at your site but on rented servers at an EU provider that we operate for you, self-hosting with Hetzner and Coolify is the right option.

It is worth it when strict rules govern where data may be, when you already have your own infrastructure, or when systems must run without internet access. It is not worth it when nobody can take over ongoing operations – updates, backups and monitoring. Then a managed cloud in the EU is often the safer choice.

On-premises, EU cloud or SaaS?

	On-premises	EU cloud	SaaS
Data location	your servers	the provider's data center in the EU	at the SaaS provider
Operation	you or a contracted service provider	shared: provider for the platform, you for the application	entirely by the provider
Scaling	limited by your own hardware	flexible	flexible, within the product
Cost structure	hardware and operation	usage-based	subscription
Suited for	strict data location rules, offline operation	custom applications without your own data center	standard processes without customization

What We Build with On-Premise

Docker-Based Deployments Ollama for Local Language Models Private Container Registries Reverse Proxy & TLS (Nginx / Traefik) Automated Updates & Rollbacks Backup & Disaster Recovery Monitoring & Alerting VPN & Network Security Hybrid Cloud Architectures Documentation & Runbooks

Use Cases

Typical Use Cases

AI Without the Cloud

Local language models with Ollama for applications with confidential data – requests and answers stay in your network.

Applications with Strict Requirements

Operation in your own infrastructure when regulation or internal policies set where data may be.

Hybrid Architectures

Sensitive data stays on your own servers, while compute-heavy or public parts run in an EU cloud.

References

Selected Projects

Web Application

Klanglicht - Art Festival of Bühnen Graz

Klanglicht Audio Guide

Mobile App

Radio Soundportal

Soundportal App

Web Application

Österreichische Raiffeisen-Sicherungseinrichtung eGen (ÖRS)

Deposit Insurance Payout Platform

View all references

FAQ

On-Premise FAQ

Can dectria deploy AI solutions on premises?

Yes. With Ollama, we run open language models such as Llama, Mistral or Gemma directly on your hardware. Requests and answers never leave your network. Local models are smaller than the large cloud models – so we test with your real tasks beforehand whether the quality is sufficient.

What hardware does a local language model need?

That depends mainly on the model size. Smaller models run on a workstation with a graphics card or on an Apple Silicon machine with a lot of memory; larger models need server GPUs. We size the hardware with a test on your real requests instead of estimating across the board.

How does dectria deliver on-premises software?

As Docker images through a private registry or by direct handover. The containers include all dependencies and run on any system with Docker. We roll out updates through automated pipelines or controlled rollouts with rollback.

Does dectria offer ongoing support for on-premises installations?

Yes. We handle security updates, monitoring, backup checks and support for extensions. We agree up front who takes on which task in operation.

Every project starts with a conversation.

Let us talk about your individual needs and goals.

Start a project